Congestion management mechanism

Abstract

A system to facilitate resolving traffic congestion in a network switching fabric is described. The system includes a processor and a machine readable medium storing instructions that, when executed, cause the processor to receive a notification from a virtual connect switch indicating that a congestion condition has been detected in remote direct memory access (RDMA) over Converged Ethernet (RoCE) data traffic flowing from a first end point to a second end point via a first of a plurality of paths between the first end point and the second end point, identify a second of the plurality of paths between the first end point and the second end point and modify the RoCE data traffic to flow from the first end point to the second end point via the second path.

What is claimed is: 
     
       1. A system to facilitate a resolution of traffic congestion in a network switching fabric, comprising:
 a processor; and 
 a non-transitory machine-readable medium storing fabric manager instructions executable on the processor to:
 receive an explicit congestion notification (ECN) from a virtual connect switch indicating that a congestion condition has been detected in remote direct memory access (RDMA) over Converged Ethernet (RoCE) internet layer data packet traffic flowing from a first end point to a second end point via a first path of a plurality of paths between the first end point and the second end point; 
 identify a second path of the plurality of paths between the first end point and the second end point, wherein the first path is through a first switch virtual interface of the virtual connect switch, and the second path is through a second switch virtual interface of the virtual connect switch; and 
 modify a route of the RoCE internet layer data packet traffic to flow from the first end point to the second end point via the second path. 
 
 
     
     
       2. The system of  claim 1 , wherein the fabric manager instructions are executable on the processor to modify the route of the RoCE internet layer data packet traffic by programming a routing table within the virtual connect switch. 
     
     
       3. The system of  claim 1 , wherein the fabric manager instructions are executable on the processor to identify a priority associated with the first path. 
     
     
       4. The system of  claim 3 , wherein the fabric manager instructions are executable on the processor to modify the route of the RoCE internet layer data packet traffic by modifying a cost associated with the second path in accordance with the priority associated with the first path. 
     
     
       5. The system of  claim 1 , wherein the fabric manager instructions are executable on the processor to receive a second ECN indicating that the congestion condition in the first path has been alleviated. 
     
     
       6. The system of  claim 5 , wherein the ECN and the second ECN are received via a Simple Network Management Protocol (SNMP). 
     
     
       7. The system of  claim 1 , wherein the fabric manager instructions are executable on the processor to modify the route of the RoCE internet layer data packet traffic by disabling a routing protocol at the first switch virtual interface. 
     
     
       8. A method to facilitate a resolution of traffic congestion in a network switching fabric, comprising:
 receiving, at a traffic manager executed in a computer system, a notification (ECN) from a virtual connect switch indicating that a congestion condition has been detected in remote direct memory access (RDMA) over Converged Ethernet (RoCE) internet layer data packet traffic flowing from a first end point to a second end point via a first path of a plurality of paths between the first end point and the second end point; 
 identifying, by the traffic manager, a second path of the plurality of paths between the first end point and the second end point, wherein the first path is through a first switch virtual interface of the virtual connect switch, and the second path is through a second switch virtual interface of the virtual connect switch; and 
 modifying, by the traffic manager, a route of the RoCE internet layer data packet traffic to flow from the first end point to the second end point via the second path. 
 
     
     
       9. The method of  claim 8 , wherein modifying the route of the RoCE internet layer data packet traffic comprises programming a routing table within the virtual connect switch. 
     
     
       10. The method of  claim 8 , further comprising identifying a priority associated with the first path. 
     
     
       11. The method of  claim 10 , wherein modifying the route of the RoCE internet layer data packet traffic comprises modifying a cost associated with the second path in accordance with the priority associated with the first path. 
     
     
       12. The method of  claim 8 , further comprising receiving a second ECN indicating that the congestion condition in the first path has been alleviated. 
     
     
       13. The method of  claim 12 , wherein the ECN and the second ECN are received via a Simple Network Management Protocol (SNMP). 
     
     
       14. The method of  claim 8 , wherein modifying the route of the RoCE internet layer data packet traffic comprises disabling a routing protocol at the first switch virtual interface. 
     
     
       15. A non-transitory machine-readable medium storing instructions which, when executed, cause a computer system to:
 receive an explicit congestion modification (ECN) from a virtual connect switch indicating that a congestion condition has been detected in remote direct memory access (RDMA) over Converged Ethernet (RoCE) internet layer data packet traffic flowing from a first end point to a second end point via a first path of a plurality of paths between the first end point and the second end point; 
 identify a second path of the plurality of paths between the first end point and the second end point, wherein the first path is through a first switch virtual interface of the virtual connect switch, and the second path is through a second switch virtual interface of the virtual connect switch; and 
 modify a route of the RoCE internet layer data packet traffic to flow from the first end point to the second end point via the second path. 
 
     
     
       16. The non-transitory machine-readable medium of  claim 15 , wherein the instructions when executed cause the computer system to modify the route of the RoCE internet layer data packet traffic by programming a routing table within the virtual connect switch. 
     
     
       17. The non-transitory machine-readable medium of  claim 15 , wherein the instructions when executed cause the computer system to identify a priority associated with the first path. 
     
     
       18. The non-transitory machine-readable medium of  claim 17 , wherein the instructions when executed cause the computer system to modify the route of the RoCE internet layer data packet traffic by modifying a cost associated with the second path in accordance with the priority associated with the first path. 
     
     
       19. The non-transitory machine-readable medium of  claim 15 , wherein the instructions when executed cause the computer system to receive a second ECN indicating that the congestion condition in the first path has been alleviated. 
     
     
       20. The non-transitory machine-readable medium of  claim 19 , wherein the ECN and the second ECN are received via a Simple Network Management Protocol (SNMP).